I have been running El Capitan on my MacPro 1,1 (firmware updated to Mac Pro2,1 but SMC left untouched) with 2 Xeon X5355 SLEAG (this stepping has enhanced halt state enabled) and 24GB RAM without issues. Looking at the specs the X5365 is specified with 150W TDP while the X5355 has 120W. The symptom you observe is similar to that of regular PCs when current draw exceeds power supply capabilities.
It only takes a small current spike to trigger that. Hence you should check the stepping of your CPUs (Enhanced Halt State is available on
X5365 SLAED) as that might help staying below the critical power draw. The reason why it works on older version of OS X is that utilisation of CPU hardware has improved with newer kernel features and hence the power draw tends to be higher on El Capitan compared to let's say Lion when CPU utilisations is high. Snow Leopard is also largely 32bit, which should also result in less power draw.
I have never really done any measurement on this though, hence there might be other issues.
Have you tried with just 16GB of RAM ? Not sure though, if that strains the same part of the power supply as the CPU.
Recovery mode may also work as there are less kext launched. Safe mode could be another option to try, which has less stuff launched as well.
The other issue with the 3.0GHz version is temperature as without SMC update the fans spin slower and you might need to use smcFanControl to adjust. However that should not kick in that shortly after startup.
Thanks! I also got 5355's SLEAG running in my other mac pro. No problems. I think i don't have the 150 watt TDP 5365's. I have the SLAED ones as you linked and i think those are 120 watt TDP. Both 2006 mac pro's have efi update and smc update to 2,1.
I did a verbose start up. Recovery mode was also looping. Safe mode maybe another option which i haven't tried.
Anyway this is what Verbose mode said about errors:
panic(cpu 4 caller 0xffffff80121ce6fa): Kernel trap at 0xffffff8012183336, type 14=page fault, registers:
CR0: 0x0000000080010033, CR2: 0xffffff801900a2a8, CR3: 0x000000004d5d4000, CR4: 0x0000000000002660
RAX: 0xffffff7f80000000, RBX: 0xffffff8016a5c1c0, RCX: 0xffffff7f80000028, RDX: 0x000000009900a280
RSP: 0xffffff8873fe3cd0, RBP: 0xffffff8873fe3d00, RSI: 0x00000000592ef5cf, RDI: 0xffffff8015d0a330
R8: 0xffffff801c63f4a0, R9: 0xffffff8873fe3e90, R10: 0x000000006a8da464, R11: 0x00007fc1c2a0fe90
R12: 0xffffff801c6c1c30, R13: 0xffffff8015b07000, R14: 0x0000000000010000, R15: 0xffffff8015d0a330
RFL: 0x0000000000010202, RIP: 0xffffff8012183336, CS: 0x0000000000000008, SS: 0x0000000000000000
Fault CR2: 0xffffff801900a2a8, Error code: 0x0000000000000000, Fault CPU: 0x4, PL: 1
I could switch the cpu's and see if the problem moves. I think CPU 4 means CPU 1 core 4. So maybe it will say cpu 8 then. But i don't want to open it up again. I double checked if the cpu's were really in the middle of the holder. But maybe when fitting the heat sinks something moved. But it could also be the CPU's aren't that great.
I also have 2 spare 5355's but i'm not sure if they are matched.
And i run 16 GB of memory. 2 times 8GB. I see no errors leds when i press diagnostic or without.
Is there a cpu stress test i can do in Lion? Geekbench 2 gives a normal score but not 100% sure it uses all the cores.